J4 ›› 2010, Vol. 37 ›› Issue (4): 655-659.doi: 10.3969/j.issn.1001-2400.2010.04.013

• 研究论文 • 上一篇    下一篇

随机集粒子滤波的快速被动数据关联算法

杨柏胜;姬红兵;高小东   

  1. (西安电子科技大学 电子工程学院,陕西 西安  710071)
  • 收稿日期:2009-03-11 出版日期:2010-08-20 发布日期:2010-10-11
  • 通讯作者: 杨柏胜
  • 作者简介:杨柏胜(1980-),男,西安电子科技大学博士研究生,E-mail: tfybs@163.com.
  • 基金资助:

    国家自然科学基金资助项目(60677040,60871074)

Fast passive data association algorithm base on the random set particle filter

YANG Bai-sheng;JI Hong-bing;GAO Xiao-dong   

  1. (School of Electronic Engineering, Xidian Univ., Xi'an  710071, China)
  • Received:2009-03-11 Online:2010-08-20 Published:2010-10-11
  • Contact: YANG Bai-sheng

摘要:

针对杂波干扰环境下的被动多目标跟踪问题,将多站集中式融合方法与概率假设密度粒子滤波递归过程相结合,实现被动多目标跟踪.进一步,将概率假设密度粒子滤波递归过程并行化处理,每个目标使用单独滤波器跟踪,避免了大量粒子的聚类过程,简化算法复杂度,进而提出一种快速被动数据关联算法.实验结果表明,与传统算法相比,新算法可以在不增加额外计算负担的基础上,有效得到每个目标的航迹.特别对于目标发生交叉的情况,能很好地区分每个目标的航迹.

关键词: 被动多目标跟踪, 粒子滤波, 概率假设密度, 数据关联

Abstract:

Multiple-target tracking based on passive measurements with clutters is a difficult problem. Multiple-senor centralized fusion scheme and paticle filter(PF)implementation for the probability hypothesis density(PHD)filter are combined to tracke multiple targets effectively. Furthermore, to sovle the incapablity of the PHD filter in maintaining the integrated trajectory of each traget, the PHD recursion is implemted in parallel, where each target is tracked with a single PF. A fast data association algorithm is deducted, in which the clustering for all particles is avoided. Simulation results show that, compared with the conventional ones, the new method can keep each tracks better without additonal computational costs. Especially, when targets cross each other, the respective trajectories can be distinguished effectively.

Key words: passive multi-target tracking, particle filter, probability hypothesis density(PHD), data association